A Novel Technique for Retrieving Source Code Duplication
Yoshihisa Udagawa · International Conference on Systems · 2014
In this paper we propose a new approach for the detection of clones in source code for improving safety of software systems. The main contributions of this paper are development of a mining algorithm to explore program structure and the definition of a similarity measure that is tailored to sequentially structured texts for retrieving similar source code fragments. Retrieval experiments were conducted using Apache-Tomcat 7, which is a large-size open source Java program. The results show that the proposed mining algorithm finds a set of frequent sequences within one minute, and the proposed similarity measure is a better indicator than the Sorensen-Dice index.